{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VFFGSMULKVLBK65A4UZ25TY23R","short_pith_number":"pith:VFFGSMUL","schema_version":"1.0","canonical_sha256":"a94a69328b5556157ba0e533aecf1adc7b51c1e6b9c581f68cf55124803ed2ab","source":{"kind":"arxiv","id":"2606.08854","version":1},"attestation_state":"computed","paper":{"title":"sGPO: Trading Inference FLOPs for Training Efficiency in RLVR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Akash Srivastava, Giorgio Giannone, Kai Xu, Shivchander Sudalairaj","submitted_at":"2026-06-07T21:47:31Z","abstract_excerpt":"Standard Reinforcement Learning with Verifiable Rewards (RLVR) training allocates a fixed rollout budget to every query, without regard for what each query's difficulty means for the current policy. This leads to two symmetric failure modes: easy queries produce near-zero advantage because the policy already solves them, while unsolvable queries produce no signal because the policy never solves them. Both regimes waste training FLOPs without contributing to a learning gradient. We introduce sorted Group Policy Optimization (sGPO), a compute-efficient strategy that trades a small budget of infe"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.08854","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-07T21:47:31Z","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"title_canon_sha256":"039fea9f6290d9123d1069bd4df344b2a2e06231f7f5f1516c441d36108dd921","abstract_canon_sha256":"231ba2bb3b0cf8dbf57fef77d63342f80eb7d9bf62769de73bbd744da7b24787"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:07:43.036959Z","signature_b64":"WrmAUhXs/o8Fh71/Dz+QuzetLtP5GoqAJJFtzNYtgaNm+eKKV/Uq5RjSQrOglmbQtSgjvZd9QGKraYbQxKeTDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a94a69328b5556157ba0e533aecf1adc7b51c1e6b9c581f68cf55124803ed2ab","last_reissued_at":"2026-06-09T02:07:43.036065Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:07:43.036065Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"sGPO: Trading Inference FLOPs for Training Efficiency in RLVR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Akash Srivastava, Giorgio Giannone, Kai Xu, Shivchander Sudalairaj","submitted_at":"2026-06-07T21:47:31Z","abstract_excerpt":"Standard Reinforcement Learning with Verifiable Rewards (RLVR) training allocates a fixed rollout budget to every query, without regard for what each query's difficulty means for the current policy. This leads to two symmetric failure modes: easy queries produce near-zero advantage because the policy already solves them, while unsolvable queries produce no signal because the policy never solves them. Both regimes waste training FLOPs without contributing to a learning gradient. We introduce sorted Group Policy Optimization (sGPO), a compute-efficient strategy that trades a small budget of infe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08854","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2606.08854/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.08854","created_at":"2026-06-09T02:07:43.036215+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.08854v1","created_at":"2026-06-09T02:07:43.036215+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08854","created_at":"2026-06-09T02:07:43.036215+00:00"},{"alias_kind":"pith_short_12","alias_value":"VFFGSMULKVLB","created_at":"2026-06-09T02:07:43.036215+00:00"},{"alias_kind":"pith_short_16","alias_value":"VFFGSMULKVLBK65A","created_at":"2026-06-09T02:07:43.036215+00:00"},{"alias_kind":"pith_short_8","alias_value":"VFFGSMUL","created_at":"2026-06-09T02:07:43.036215+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R","json":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R.json","graph_json":"https://pith.science/api/pith-number/VFFGSMULKVLBK65A4UZ25TY23R/graph.json","events_json":"https://pith.science/api/pith-number/VFFGSMULKVLBK65A4UZ25TY23R/events.json","paper":"https://pith.science/paper/VFFGSMUL"},"agent_actions":{"view_html":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R","download_json":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R.json","view_paper":"https://pith.science/paper/VFFGSMUL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.08854&json=true","fetch_graph":"https://pith.science/api/pith-number/VFFGSMULKVLBK65A4UZ25TY23R/graph.json","fetch_events":"https://pith.science/api/pith-number/VFFGSMULKVLBK65A4UZ25TY23R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R/action/storage_attestation","attest_author":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R/action/author_attestation","sign_citation":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R/action/citation_signature","submit_replication":"https://pith.science/pith/VFFGSMULKVLBK65A4UZ25TY23R/action/replication_record"}},"created_at":"2026-06-09T02:07:43.036215+00:00","updated_at":"2026-06-09T02:07:43.036215+00:00"}